Everyone should learn GitHub, even if you never write code

GitHub is a website that keeps every saved version of a project, so you can see what changed, go back to any earlier point, and share the work with others. It was built for programmers. I think everyone who uses AI to make things should learn it, because AI makes changes fast, and GitHub is the best undo button there is.

The problem GitHub solves

You’ve seen folders like this:

report.docx
report-final.docx
report-final-v2.docx
report-final-v2-REAL.docx

That’s version control done by hand. It breaks down the moment a project has more than one file, or more than one person, or an AI assistant that edits fourteen files in one go.

Git, the tool underneath GitHub, does this properly. Every time you save a version, it records exactly what changed across the whole project, with a short note saying why. You can compare any two versions, or go back to any of them.

Why AI makes this urgent

Before AI, a non-programmer rarely changed many files at once. Now you can ask an assistant to “redo the pricing section” and it will touch your page, your styles and your data in seconds. Sometimes that’s great. Sometimes it quietly breaks something that worked yesterday.

With versions saved on GitHub:

• A bad change costs seconds. You go back to the last good version.
• You can see what the AI did. GitHub shows the changes line by line, in green and red.
• Nothing is lost if your laptop dies. The project lives online too.

In my own work, the rule is to save a version before every request to an AI. It’s the first of the habits I recommend for building with AI.

The words, in plain language

You only need six to start.

Word	What it means
Repository (repo)	A project folder that remembers its whole history
Commit	One saved version, with a note about what changed
Branch	A separate copy to try something without touching the main one
Push	Send your saved versions from your computer up to GitHub
Pull request	“Here are my changes, please look before they go in”
Merge	Bring the changes from a branch into the main version

If Git and GitHub sound like the same thing, they aren’t quite. I explain the difference in Git vs GitHub.

It’s not only for code

A repository can hold any files that are mostly text:

• Writing. Books, scripts, blog posts. This blog is a folder of text files on GitHub. Publishing a post means saving a file.
• Notes and documents. Policies, recipes, lesson plans. You get a full history of who changed what.
• Websites. Many hosts publish a site straight from a repository. Save a change, and the site updates a minute later.
• Instructions for AI. The files that tell an assistant how your project works live in the repo too.

AI tools already live there

This is the part that convinced me everyone needs it. AI coding tools don’t only write files on your computer. Increasingly they work through GitHub directly: they open pull requests, review changes, and run checks. Hosting services watch your repository and publish whenever it changes.

That’s how I put projects online. Each one is a repository. When I save a change, the host notices and deploys it, which I describe in shipping solo with AI. If the new version is broken, I go back to the previous one with a click.

Learning GitHub is learning the place where AI work actually happens.

A plan for your first hour

You don’t need the command line to begin.

1. Create a free account at github.com.
2. Make a repository with the “New” button. Tick “Add a README file”.
3. Edit the README in the browser. Click the pencil icon, type a few lines, and press “Commit changes”. You’ve made a commit.
4. Look at the history. Click “Commits” and open one. The green and red lines are what changed.
5. Upload a file. “Add file”, then “Upload files”. Commit again.
6. Install GitHub Desktop when you’re ready to work on your own computer. It gives you buttons for commit and push, with no terminal.

Then, the next time you ask an AI to change something, commit first. That one habit is most of the value.

Common worries

“Is my work public?” Only if you choose. Repositories can be private, and private is free for individuals.

“Will I break something?” It’s hard to. The whole point is that earlier versions stay safe.

“Do I need to learn commands?” Not at first. The website and the desktop app cover the basics. An AI assistant can also run the commands for you, and explain each one if you ask.

“Is this overkill for a small project?” A repository costs nothing and takes a minute. The first time it saves you from a bad change, it has paid for itself.

Where to go next

Once you’ve made a few commits, branches and pull requests will make sense quickly, because they’re just safer ways to try things. If you’re building with AI, pair this with a short rules file for your assistant, which I cover in how to write a CLAUDE.md.